JP6482959B2 - Semiconductor memory device - Google Patents

Description

  The present invention relates to a memory device, and in particular, is a technique applicable to a semiconductor memory device using a resistance change element.

  One type of nonvolatile memory is a resistance change memory (ReRAM) that utilizes a difference in resistance value between a low resistance state (On state) and a high resistance state (Off state) of a resistance change element.

  As a technique for improving the stability and reliability of the write operation in ReRAM, memory retention characteristics, and the like, for example, in Patent Document 1 (Japanese Patent No. 4838399), the first resistance state is applied when a pulse of the first voltage is applied. A resistance variable element having a characteristic that changes from a second resistance state to a first resistance state when a pulse of a second voltage having a polarity different from that of the first voltage is applied. Thus, when changing from the first resistance state to the second resistance state, at least a pulse of the first voltage and a third voltage having an absolute value smaller than the second voltage and having the same polarity as the second voltage. And a method of applying a first voltage pulse in this order.

  Patent Document 2 (Japanese Patent No. 5250726) includes a weak write step and a subsequent normal write step as steps for changing the resistance variable element from the first resistance state to the second resistance state. In the weak writing step, after the third resistance pulse is applied to the resistance variable element by applying a pulse of a third voltage having the same polarity as the first voltage but different in absolute value, the absolute voltage having the same polarity as the second voltage is applied. Transition to an intermediate resistance state having a resistance value between the first resistance state and the second resistance state by applying a pulse of a fourth voltage whose value is smaller than the second voltage and the third voltage, A technique is described in which a transition from the intermediate resistance state to the second resistance state is performed by applying a pulse of the first voltage to the resistance variable element at least once.

Japanese Patent No. 4838399 Japanese Patent No. 5250726

  In order to increase the memory window (ratio of the On resistance that is the resistance value of the On state and the Off resistance that is the resistance value of the Off state) in order to improve the performance in the ReRAM, the writing in the Off state (Off writing) It is effective to increase the Off resistance by increasing the conditions (for example, increasing the amplitude of the pulse voltage applied during Off writing, increasing the pulse width, increasing the pulse current, etc.). . However, when the off-write condition is excessively increased, the on-state retention characteristic when the state is subsequently changed to the on-state tends to deteriorate. That is, there is a trade-off relationship between securing the memory window and holding characteristics of the On state.

  In this regard, for example, in the technique described in Patent Document 1, for the purpose of improving the stability of resistance after writing and improving the verification success rate, it is superimposed on the first pulse of the first voltage. Before applying the second pulse of the first voltage, the third voltage pulse having the opposite polarity is applied. On the other hand, considering the retention characteristics after writing in this case, it becomes necessary to lengthen the pulse of the first voltage, and as a result, a long pulse is repeatedly applied, so that the writing time and writing energy increase. It has the problem of being able to.

  Further, for example, in the technique described in Patent Document 2, the third voltage pulse applied in the weak write step and the first voltage pulse of the same polarity applied in the normal write step are used from the viewpoint of power consumption reduction. The absolute value of the third voltage is mainly assumed to be smaller than the first voltage. In this case, for example, when the resistance is lowered (on writing), the probability and degree of the low resistance state (the state in which the conductive filament is connected in the resistance change element) is reduced, so that the repair function of the retention failure is achieved. It has the subject that it may fall and sufficient effect may not be acquired.

  That is, an object of the present application is to improve the reliability of the semiconductor memory device. Other problems and novel features will become apparent from the description of the specification and the accompanying drawings.

  A semiconductor memory device according to an embodiment includes: a memory cell including a resistance change element; and a first resistance value of the resistance change element that is less than a first reference value relative to the memory cell. A first write process for applying a first write pulse to achieve a resistance state of the second and a second resistance state having a polarity opposite to that of the first write pulse to obtain a second resistance state equal to or greater than a second reference value. And a control circuit capable of performing a second writing process in which two writing pulses are applied. The control circuit has the same polarity as the first write pulse and a pulse from the first write pulse before applying the first write pulse to the memory cell in the first write process. A first pulse having a short width and a second pulse having the same polarity as the second write pulse are applied in this order.

  According to the above embodiment, the reliability of the semiconductor memory device can be improved. In particular, in a semiconductor memory device composed of resistance change elements, it is possible to improve on-state retention characteristics while securing a memory window.

It is the figure which showed the outline | summary about the structural example of the resistance change element used with bipolar type ReRAM. It is the figure which showed the outline | summary about the structural example of the memory cell. It is the figure which showed the outline | summary about the structural example of the memory cell array in ReRAM. It is the figure which showed the example of a waveform of each normal applied voltage at the time of performing the write which makes the memory cell an On state, and the write which makes an Off state. It is the figure which showed the example of a waveform of each applied voltage at the time of writing which makes the memory cell in Embodiment 1 of this invention the write which makes an On state, and an Off state. It is the figure which showed the specific example of the sequence of writing and reading in Embodiment 1 of this invention. It is the figure which showed the example of the distribution condition of the On resistance and the Off resistance in the case where it is based on the case where it is based on the on-writing system in Embodiment 1 of this invention, and a conventional system. It is the figure which showed the example of the retention characteristic of the On state in the case where it is based on the case where it is based on the case of the On writing system in Embodiment 1 of this invention and a conventional system. It is the figure which showed the example of the holding | maintenance characteristic of the On state at the time of changing the pulse width of a trial pulse and a main pulse in the On writing system in Embodiment 1 of this invention. It is the figure which showed the example of the retention characteristic of the On state at the time of changing the voltage of a trial pulse in the On writing system in Embodiment 2 of this invention. It is the figure which showed the outline | summary about the example of operation | movement and the effect of a resistance change element in Embodiment 2 of this invention. (A), (b) is the figure which showed the outline | summary about the example of operation | movement and an effect of a resistance change element at the time of changing the conditions of the trial pulse in Embodiment 2 of this invention. It is the figure which showed the example of the holding | maintenance characteristic of the On state at the time of changing the repetition frequency of the sequence and trial pulse of the On write system in Embodiment 3 of this invention, and a reset pulse. It is the figure which showed the example of the sequence of the On write system in Embodiment 4 of this invention. It is the figure which showed the example of a waveform of each applied voltage at the time of performing On write and Off write in Embodiment 5 of this invention. It is the figure which showed the outline | summary about the example of the structure and operation | movement of a resistance change element in Embodiment 6 of this invention. It is the figure which showed the outline | summary about the structural example of the memory cell of crosspoint type | mold ReRAM. It is the figure which showed the outline | summary about the structural example of the memory cell array in crosspoint type | mold ReRAM.

  Hereinafter, embodiments will be described in detail with reference to the drawings. Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.

  As described above, in writing to the ReRAM, there is a trade-off relationship between securing the memory window and holding characteristics of the On state. Here, the inventors of the present application first applied a short On pulse at the time of On writing, and then had the same polarity as that of the pulse at Off writing (that is, the polarity opposite to the On pulse). It has been found that the on-state retention characteristic can be improved while securing the memory window by applying a sequence of applying a pulse having a small value and then applying a longer On pulse.

  Therefore, in the embodiment shown below, a resistance change type memory using a resistance change element as a nonvolatile memory element, in particular, pulses with different polarities are applied in low resistance write (On write) and high resistance write (Off write). When the On-write is performed on the bipolar type ReRAM, the above sequence is taken to improve the On-state retention characteristics while securing the memory window, thereby improving the ReRAM performance.

(Embodiment 1)
FIG. 1 is a diagram showing an outline of a structural example of a variable resistance element used in a bipolar ReRAM. The resistance change element VR has a configuration in which the resistance change layer VRL is sandwiched between the metal layer M1 and the metal layer M2, and the metal layer M1 and the metal layer M2 form a first electrode and a second electrode, respectively. ing. The resistance change layer VRL is changed to a low resistance state (On state) by applying a positive voltage to the metal layer M2 with reference to the metal layer M1, and a positive voltage is applied to the metal layer M1 with reference to the metal layer M2. Thus, the resistance change layer VRL can be changed to the high resistance state (Off state). One-bit information is stored by making the On state and Off state correspond to 0 and 1 or 1 and 0, respectively.

  The resistance change layer VRL is formed of, for example, a metal oxide (for example, tantalum oxide, titanium oxide, zirconium oxide, or hafnium oxide). In this case, the resistance change layer VRL may be a single layer film or a laminated film. When the resistance change layer VRL is a laminated film, the resistance change layer VRL may be, for example, a laminated film having a different combination of element types, or a laminated film having the same combination of element types. Also good. In this case, the oxygen composition ratios of the layers of the laminated film are different from each other. The film thickness of the resistance change layer VRL is, for example, not less than 1.5 nm and not more than 30 nm. The metal layer M1 and the metal layer M2 are each formed of, for example, ruthenium, titanium nitride, tantalum, tantalum nitride, tungsten, palladium, or platinum.

  FIG. 2 is a diagram showing an outline of a configuration example of a memory cell in the ReRAM. The memory cell MC can be configured by combining the resistance change element VR shown in FIG. 1 and a selection transistor TR formed of a MOS (Metal-Oxide Semiconductor) transistor. The selection transistor TR is a selection transistor that controls whether a potential difference between the bit line BL and the plate line PL is applied to or cut off from the resistance change element VR.

  The resistance change element VR has one terminal connected to the plate line PL, the other terminal connected to the bit line BL via the selection transistor TR, and the gate of the selection transistor TR connected to the word line WL. The polarity of the voltage applied to the resistance change element VR can be switched depending on which of the potential of the bit line BL and the potential of the plate line PL is higher than the other.

  Which of the metal layer M1 and the metal layer M2 is connected to the bit line BL is not particularly limited, but in the following description, it is assumed that the metal layer M1 is connected to the bit line BL. In addition, the selection transistor TR is not limited to the N-channel type or the P-channel type, but in the following, it is assumed that the selection transistor TR is an N-channel type in which a source and a drain are electrically connected by applying a positive voltage to the gate. explain. In the case of the P-channel type, the source and the drain are made conductive by applying a negative voltage to the gate.

  FIG. 3 is a diagram showing an outline of a configuration example of a memory cell array in the ReRAM. The memory cell array MCA can be configured by arranging the memory cells MC shown in FIG. 2 in a matrix. The example of the memory cell array MCA shown in FIG. 3 has a 16-bit storage capacity composed of a matrix of 4 rows × 4 columns, but a larger storage capacity is realized by appropriately increasing the number of rows and columns in the array. be able to.

  Each memory cell MC is connected to each intersection of word lines WL0 to WL3, bit lines BL0 to BL3, and plate lines PL0 to PL3. All word lines WL0 to WL3, bit lines BL0 to BL3, and plate lines PL0 to PL3 are connected to a control circuit (not shown) in the periphery of the memory cell array MCA. For example, the word lines WL0 to WL3 are connected to a word line control circuit (not shown) on the left side of the figure in the memory cell array MCA. The bit lines BL0 to BL3 are connected to a bit line control circuit (not shown) in the upper part of the drawing. Similarly, the plate lines PL0 to PL3 are connected to a plate line control circuit (not shown) at the top in the drawing.

  Each control circuit performs writing by appropriately applying a voltage to the word line WL, the bit line BL, and the plate line PL to bring a desired memory cell MC into a high resistance state or a low resistance state. Alternatively, reading is performed by detecting a current flowing through the bit line BL or the plate line PL and determining whether a desired memory cell MC is in a high resistance state or a low resistance state.

  For example, in writing to turn on the memory cell MC surrounded by a dotted circle, the word line WL1 and the plate line PL1 are set to a high potential, and the other word lines WL0, WL2, WL3, and plate lines PL0, PL2, PL3. All the bit lines BL0 to BL3 may be set to zero potential. On the other hand, in writing to turn off the memory cell MC surrounded by the dotted circle, the word line WL1 and the bit line BL1 are set to a high potential, and the other word lines WL0, WL2, WL3, and the bit lines BL0, BL2, BL3 and all the plate lines PL0 to PL3 may be set to zero potential.

  In addition, in order to read whether the memory cell MC surrounded by the dotted circle is in the On state or the Off state, the word lines WL0, WL2, WL3 other than the word line WL1 and the plate line PL1, and the plate lines PL0, PL2, PL3, All the bit lines BL0 to BL3 are set to zero potential, and the word line WL1 is set to high potential. Then, a voltage sufficiently lower than that at the time of writing may be applied to the plate line PL1, and the current flowing through the bit line BL1 or the plate line PL1 may be detected.

  In the above operation, in the memory cell MC connected to other than the word line WL1, the selection transistor TR becomes non-conductive and no voltage is applied to the resistance change element VR. Further, in the memory cell MC connected to other than the bit line BL1 and the plate line PL1, no voltage is applied to the resistance change element VR because the bit lines BL0, BL2, BL3 and the plate lines PL0, PL2, PL3 have the same potential. . As a result, only the memory cell MC surrounded by the dotted circle is written or read. In writing and reading to other memory cells MC, writing and reading can be performed by the same method.

  FIG. 4 is a diagram showing waveform examples of normal applied voltages at the time of performing writing to turn on and turn off the memory cell MC. In order to set the resistance change element VR of the memory cell MC shown in FIG. 2 to a low resistance state (On state), a voltage higher than that on the bit line BL side is applied to the plate line PL side of the resistance change element VR. Is applied once in the form of a pulse as shown on the left side of FIG. For this purpose, for example, after the potential on the plate line PL side is higher than the potential on the bit line BL (in the example of FIG. 4, the PL side is set to Von and the BL side is set to zero potential), a predetermined period, The selection transistor TR may be made conductive by increasing the potential of the word line WL. Alternatively, a pulse voltage having a positive potential on the plate line PL side may be applied between the plate line PL and the bit line BL in a state where the potential of the word line WL is increased to make the selection transistor TR conductive.

  Conversely, in order to set the resistance change element VR to the high resistance state (off state), a voltage higher than the plate line PL side on the bit line BL side of the resistance change element VR is pulsed as shown on the right side of FIG. It is applied once (shown as a pulse in the reverse direction since the applied voltage has a reverse polarity with respect to the pulse for turning on). For this purpose, for example, the potential on the bit line BL side is set higher than the potential on the plate line PL (in the example of FIG. 4, the bit line BL side is set to Voff and the plate line PL side is set to zero potential). For a predetermined period, the selection transistor TR may be turned on by increasing the potential of the word line WL. Alternatively, a pulse voltage having a positive potential on the bit line BL side may be applied between the bit line BL and the plate line PL in a state where the potential of the word line WL is increased to make the selection transistor TR conductive.

  FIG. 5 is a diagram illustrating waveform examples of applied voltages when writing to set the memory cell MC in the On state and writing to set the Off state in the first embodiment. In the example of FIG. 5, the example of the applied waveform for the right off writing is the same as the normal one shown in FIG. 4. In the example of FIG. 5, it is shown that the normal applied voltage at the time of Off writing is Voff1, and the application time is Toff1.

  In the example of the left On-write applied waveform, unlike the example shown in FIG. 4, before applying a normal On-write pulse (hereinafter sometimes referred to as “Main pulse”, “main pulse”, etc.). Apply an On pulse having the same polarity as the main pulse and a short pulse width (hereinafter sometimes referred to as “Trial pulse”, “trial pulse”, etc.), and then a pulse having the opposite polarity to the On pulse. (Hereinafter, it may be described as “Recovery pulse”, “Reset pulse”, etc.).

  Here, the application time (pulse width) Ton2 of the trial pulse in the drawing is shorter than the application time Ton1 of the main pulse. That is, it has a relationship of Ton2 <Ton1. Further, the applied voltage Von2 of the trial pulse is the same as or larger than the applied voltage Von1 of the main pulse. That is, there is a relationship of | Von2 | ≧ | Von1 |. Furthermore, it is assumed that the reset pulse application time Toff2 is shorter than the application time Toff1 in the normal Off write pulse in the right diagram. That is, there is a relationship of Toff2 <Toff1. The applied voltage Voff2 of the reset pulse is smaller than the applied voltage Voff1 in the normal off write pulse. That is, there is a relationship of | Voff2 | <| Voff1 |.

  FIG. 6 is a diagram showing a specific example of the writing and reading sequences in the first embodiment. Here, before applying the main pulse (Main pulse) of 2.5 V / 2 us in the on-write sequence, first, a trial pulse (Trial pulse) having the same polarity as the main pulse and a short pulse width of 100 ns is applied. Further, it is shown that a 1.5 V / 20 ns reset pulse (Recovery pulse) having a polarity opposite to that of the main pulse is applied. The Off write pulse is 2.5 V / 100 ns, and the applied voltage of the read pulse is 0.5 V.

  FIG. 7 is a diagram showing an example of the distribution state of the On resistance and the Off resistance in the case of using the On writing method and the conventional method in the present embodiment. Here, the cumulative frequency distribution of the On resistance value and the Off resistance value evaluated in the 2 Mb memory cell array is shown for each of the On write sequence shown in FIG. 6 and the On write in which only the main pulse is applied according to the conventional method. As shown in the figure, even when the on-write sequence in the present embodiment is taken, the same resistance distribution as in the conventional method is obtained. That is, a similar memory window (ratio of On resistance to Off resistance) is obtained.

  FIG. 8 is a diagram showing an example of on-state retention characteristics in the case of using the on-write method and the conventional method in the present embodiment. Here, the time dependency of retention failure at 200 ° C. in the On state evaluated by the 2 Mb memory cell array is shown for each of the On write sequence shown in FIG. 6 and On write in which only the main pulse is applied according to the conventional method. A bit whose On resistance exceeds 30 kΩ is regarded as a defective holding bit. From FIG. 8, it can be seen that the on-state retention characteristic is improved by the on-write method in the present embodiment. That is, considering the contents of FIG. 7 together, it can be seen that the on-state retention characteristic is improved without affecting the memory window by the on-write method according to the present embodiment.

  FIG. 9 is a diagram showing an example of on-state retention characteristics when the pulse widths of the trial pulse and the main pulse are changed in the on-writing method according to the present embodiment. Here, in the on-write sequence as shown in the upper diagram, when the sum (Tpl1 + Tpl2) of the trial pulse (Trial pulse) width (Tpl1) and the main pulse (Main pulse) width (Tpl2) is changed The retention failure bit rate at 200 ° C. in the On state evaluated in the 2Mb memory cell array is shown in the lower graph. As in the case of FIG. 8 described above, a bit whose On resistance exceeds 30 kΩ is regarded as a defective holding bit.

  In FIG. 9, Tpl2 that is the width of the main pulse is set to three types of 0.5 us, 1.0 us, and 2.0 us, and the retention defective bit rate when Tpl1 that is the width of the trial pulse is changed for each of them. Tpl1 dependence is shown. Further, the retention defective bit rate with respect to the pulse width (Tpl2) when only the main pulse is applied alone as a reference (ref.) Is also shown.

  From FIG. 9, regarding the Tpl2 dependency, if the main pulse is shortened to 0.5 us, the On state retention failure bit rate increases. However, until Tpl2 is about 1.0 us, the On state retention failure occurs even if the time is shortened. It can be seen that the bit rate does not increase. If Tpl2 is up to about 1.0 us, the dependency of the retention failure bit rate on Tpl1 is small (the retention failure bit rate does not change much even if Tpl1 changes), and Tpl1 can be shortened to 10 ns. I understand.

  When this is compared with the reference (ref.) Of the conventional method, in the reference, the retention defective bit rate increases when Tpl2 is shortened (for example, when it is shortened to 1.5 us, it exceeds 0.01). On the other hand, in the on-write method of the present embodiment, even if the main pulse Tpl2 is shortened to 1 us, the on-state retention defective bit rate can be suppressed to about 0.002 to 0.003. That is, the retention defective bit rate can be reduced with a shorter pulse width compared to the conventional method in which a single main pulse (On write pulse) is applied, and the energy required for On write can be reduced.

  As described above, according to the ReRAM of the first embodiment, the on-state retention characteristic can be improved without affecting the memory window by taking the above-described on-write sequence. Further, compared to the conventional method in which a single main pulse (On write pulse) is applied, the energy required for On write can be reduced, that is, On write with low power consumption can be achieved.

(Embodiment 2)
In the present embodiment, the application voltage (Vpl1) of the trial pulse is optimized in the On write sequence in the ReRAM of the first embodiment. FIG. 10 is a diagram illustrating an example of an on state retention characteristic when the trial pulse voltage is changed in the on write method according to the present embodiment. Here, Tpl1, which is the width of the trial pulse, is 2 ns of 20 ns and 100 ns, and the trial pulse voltage (Vpl1) is changed to 1.8 V, 2.2 V, and 2.5 V for 2 Mb, respectively. The Vpl1 dependency of the retention failure bit rate at 200 ° C. in the On state evaluated by the memory cell array is shown. As in the case of FIGS. 8 and 9 described above, a bit having an On resistance exceeding 30 kΩ is regarded as a defective retention bit.

  As shown in FIG. 10, it can be seen that, in both cases where Tpl1 is 20 ns and 100 ns, the higher the value of Vpl1, the lower the retention defective bit rate in the On state, and the better the retention characteristics. Therefore, in addition to the trial pulse desirably having a shorter width (shorter time) than the main pulse as described in the first embodiment, it is desirable that the trial pulse has a high voltage.

  As described above, in the on-write sequence of the present embodiment, since the on-state retention characteristics are improved by adding the trial pulse and the reset pulse before the main pulse, such a pulse is used. It is clear that the sequence has the effect of suppressing the On state retention failure. Therefore, in the following, the reason why it is desirable that the trial pulse is a high voltage will be considered.

  FIG. 11 is a diagram showing an outline of an example of the operation and effect of the resistance change element VR in the second embodiment. In the figure, in the structure of the resistance change element VR as shown in FIG. 1, the conductive filament FI formed in the resistance change layer VRL when the trial pulse → the reset pulse → the main pulse is applied in order from the top. An example of the state is shown.

  The conductive filament FI is formed by a state in which defects lacking oxygen (oxygen vacancies (Vo)) are gathered at a high density. When the conductive filament FI is in a state of connecting the metal layer M1 and the metal layer M2 (for example, the upper or lower diagram in FIG. 11), the resistance change element VR is in a low resistance state (On state). On the other hand, when the conductive filament FI does not completely connect the metal layer M1 and the metal layer M2 (there is a gap between them) (for example, the middle diagram in FIG. 11), the resistance change element VR has a high resistance. It becomes a state.

  As in the conventional method, when the normal main pulse is applied only once in the on-writing, the conductive filament FI is thin, for example, as shown in the upper diagram of FIG. 11, or the density of oxygen vacancies is low. There are cases where the state holding characteristics are poor (holding failure).

  In such a case, in the present embodiment, a reset pulse having a reverse polarity is once applied to the state immediately after the conductive filament FI is connected as shown in the upper diagram of FIG. Form a resistance state. As a result, when the main pulse is subsequently applied, the electric field due to the main pulse is intensively applied to the high resistance portion (gap portion of the conductive filament FI), and as a result, the lower diagram of FIG. It is considered that the conductive filament FI having a large diameter or a high density of oxygen vacancies is formed as shown in FIG. And it is thought that the holding | maintenance characteristic of an On state improves with the conductive filament FI formed in this good state (high density).

  FIG. 12 is a diagram showing an outline of an example of the operation and effect of the resistance change element VR when the trial pulse condition is changed. FIG. 12A shows an example in which the pulse width of the trial pulse is increased. In this case, as shown in the upper diagram of FIG. 12 (a), even if the conductive filament FI is formed in a bad state (low density), which is likely to be a holding failure in the On state, the pulse width of the trial pulse is reduced. The conductive filament FI grows in the lateral direction (that is, the direction orthogonal to the direction in which the metal layer M1 and the metal layer M2 face each other) by applying the electrode for a long time for a long time.

  As a result, even if a reset pulse is applied thereafter, the conductive filament FI tends to remain without being completely cut, as shown in the middle diagram of FIG. In this case, even if a main pulse is applied thereafter, a sufficient electric field is not generated in the high resistance portion, and as a result, as shown in the lower diagram of FIG. It is considered that is easily formed.

  On the other hand, FIG. 12B shows an example in which the trial pulse voltage is lowered. In this case, as shown in the upper diagram of FIG. 12B, the probability that the conductive filament FI is connected after application of the trial pulse is lowered. That is, there are many bits in the high resistance state (off state) where the conductive filament FI is not connected even after the trial pulse is applied.

  In the on-write sequence of the present embodiment, a reset pulse is applied to the state immediately after the conductive filament FI is connected, so that a current flows through the conductive filament FI, and oxygen and oxygen vacancies due to the Joule heat. It is considered that the high resistance state is once formed by cutting the conductive filament FI by the reaction with. Then, as shown in the upper diagram of FIG. 12B, even if the reset pulse is applied in a state where the conductive filament FI is not connected, the current does not flow, so that only the effect caused by the electric field is generated, As shown in the middle diagram of FIG. 12B, the conductive filament FI does not change much from the state of the upper diagram, and the effect of once resetting to the high resistance state as shown in FIG. 11 can be obtained. Can not.

  Therefore, it is considered that even if the main pulse is applied thereafter, the probability that a thin or low-density conductive filament FI is formed increases as shown in the lower diagram of FIG. As described above, in the on-write sequence of the present embodiment, the trial pulse desirably has a shorter pulse width (shorter time) than the main pulse, and preferably has a higher voltage than the main pulse. Conceivable.

  As described above, according to the ReRAM of the second embodiment, the pulse width of the trial pulse is made shorter and higher than the main pulse in the on-write sequence as described above. This suppresses the expansion (transverse in the lateral direction) of the conductive filament FI after applying the trial pulse, improves the connection probability of the conductive filament FI, and effectively causes the on-state retention failure. Holding characteristics can be further improved by making the repair possible.

(Embodiment 3)
FIG. 13 is a diagram illustrating an example of an on-state retention characteristic when the on-write method sequence according to the third embodiment and the number of repetitions of the trial pulse and the reset pulse are changed. As shown in the upper diagram of FIG. 13, in this embodiment, as a modified example of the On write sequence shown in Embodiments 1 and 2, a set of trial pulses and reset pulses is set n times ( n ≧ 2) It is assumed that the main pulse is applied after applying repeatedly.

  The lower diagram in FIG. 13 shows the dependency of the retention failure rate in the On state at 200 ° C. on the number of repetitions when the number of repetitions when applying the set of trial pulse and reset pulse in the above sequence is changed. Yes. According to this, it can be seen that the on-state retention failure rate decreases as the number of repetitions when applying the set of trial pulses and reset pulses increases.

  As described above, according to the ReRAM of the third embodiment, in the on-write sequence as shown in the first and second embodiments, by repeatedly applying a set of trial pulses and reset pulses multiple times, Furthermore, the on-state retention characteristics can be improved.

(Embodiment 4)
FIG. 14 is a diagram illustrating an example of an on-write method sequence according to the fourth embodiment. In the present embodiment, so-called verify writing is combined with the on-write sequence as shown in the first to third embodiments. That is, after performing an on-write by the on-write sequence as shown in the first to third embodiments, a read (Read) pulse is applied to check whether or not the write has been performed correctly, and the bit is not successfully written By performing additional writing on (bits that have not been sufficiently reduced in resistance), the writing success rate is improved.

  The sequence for performing verification is not particularly limited, and various verification methods can be used as appropriate. For example, although not shown, after applying the read pulse to verify the write state, the On write sequence as shown in the first to third embodiments, that is, the set of the trial pulse and the reset pulse is performed n times (n ≧ 1). It is possible to repeatedly perform overwriting verification in which additional writing is performed according to a sequence in which the main pulse is applied after that until the writing is successful or a predetermined number of times is reached. Alternatively, as shown in the upper diagram of FIG. 14, after verifying the write state by applying the read pulse, overwrite verification in which additional writing is performed by applying only the main pulse, as shown in the lower diagram of FIG. 14. In addition, “swaying” verification in which a set of a reset pulse and a main pulse is applied may be repeatedly performed.

  As described above, according to the ReRAM of the fourth embodiment, the success rate of the on-write can be improved by combining the verify write with the on-write sequence as shown in the first to third embodiments. . Compared with the overwriting verification of only the main pulse as shown in the upper diagram of FIG. 14, the “swing” verification by the reset pulse and the main pulse as shown in the lower diagram of FIG. By stabilizing the On state by the FI reset effect, it is possible to further expect the improvement effect of the retention characteristics.

(Embodiment 5)
FIG. 15 is a diagram illustrating examples of waveforms of applied voltages when performing on-writing and off-writing in the fifth embodiment. In the example of FIG. 15, the example of the left on-write sequence is the same as that shown in FIG. 5 of the first embodiment. On the other hand, in the present embodiment, in the right off writing, the normal off write pulse shown in FIG. 5 is applied and then the off state stabilization pulse is further applied. Here, it is assumed that the applied voltage Voff3 of the off-state stabilization pulse is smaller than the applied voltage Voff1 of the normal off write pulse. That is, there is a relationship of | Voff3 | <| Voff1 |.

  By applying an off state stabilization pulse (weak off pulse) after application of a normal off write pulse, the off state after writing is stabilized. By performing the On write sequence as shown in Embodiments 1 to 4 on the stabilized Off state, the On write can be effectively performed, and the holding property of the On state can be improved. As shown in FIG. 15, the off-state stabilization pulse may be applied immediately after the normal off write pulse is applied, or after the normal off write pulse is applied, the first embodiment is applied. It may be applied immediately before the On write sequence as shown in FIG.

  As described above, according to the ReRAM of the fifth embodiment, after the Off write, the Off state stabilization pulse is applied before the next On write is performed, thereby stabilizing the Off state. Form. By performing the on-write sequence as shown in the first to fourth embodiments with respect to the stabilized off state, it is possible to effectively perform the on-write and improve the on-state retention characteristics. Can do.

(Embodiment 6)
In the present embodiment, an example of the element structure of the resistance change element VR that can obtain a greater effect when the on-write sequence as described in the first to fifth embodiments is applied will be described. The on-write method as shown in the first to fifth embodiments relates to a write technique that improves the stability in the on state. Therefore, a remarkable effect can be obtained when applied to a variable resistance element VR having a structure in which the On state is likely to become unstable, in other words, the structure in which the Off state is relatively more stable than the On state. Can do.

  FIG. 16 is a diagram schematically showing an example of the structure and operation of resistance change element VR according to the sixth embodiment. In the diagram on the left side of FIG. 16, On writing, that is, when the potential Vbl of the electrode (metal layer M1) on the bit line BL side is set to zero and the positive voltage Vpl is applied to the electrode (metal layer M2) on the plate line side. The situation of the resistance change element VR is shown.

  As a cause of the structure in which the On state is easily destabilized, for example, a case where the initial resistance of the resistance change element VR is low and a current easily leaks between the electrodes (“leaky”) can be considered. In this embodiment, “low” initial resistance depends on the dimensions of the ReRAM. For example, the resistivity of the resistance change layer VRL indicates 1e6 Ω · cm or less. In this case, a remarkable effect can be obtained by applying the on write method as shown in the first to fifth embodiments.

  When the initial resistance is low, there are a plurality of current paths (indicated by arrows in the figure) as shown in the upper left diagram of FIG. 16 at the time of forming and on-writing to form a current conduction path (path). Therefore, a sufficient current is not supplied to the portion where the conductive filament FI is to be formed, and the conductive filament FI having a low density of oxygen vacancies Vo (white circles in the drawing) is easily formed.

  As a case where the initial resistance of the resistance change element VR is low, for example, a case where the oxygen vacancies Vo are introduced in advance in the resistance change layer VRL in the initial state before forming is considered. Specifically, for example, there is a case where the metal layer M2 (that is, the electrode to which a positive voltage is applied during On writing) includes a metal that has an effect of extracting oxygen from the resistance change layer VRL. . Whether or not the metal layer M2 has an effect of extracting oxygen from the resistance change layer VRL is determined by the material used for the metal layer M2 and the physical properties of the material of the resistance change layer VRL. That is, when the material used for the metal layer M2 has a characteristic that it is more likely to react with oxygen than the material of the resistance change layer VRL, the metal layer M2 has an effect of extracting oxygen from the resistance change layer VRL.

  On the other hand, in the diagram on the right side of FIG. 16, Off writing, that is, the potential Vpl of the electrode (metal layer M2) on the plate line PL side is set to zero, and the positive voltage Vbl is applied to the electrode (metal layer M1) on the bit line side. The situation of the variable resistance element VR is shown.

  Specifically, as a structure in which the Off state is more easily stabilized, for example, the material forming the metal layer M1 (that is, the electrode on the side to which a positive voltage is applied during Off writing) is ruthenium, platinum, gold, Examples include noble metals that do not react with oxygen, such as iridium. In such a structure, when Off writing is performed, oxygen O (black circles in the figure) is likely to be accumulated near the interface between the resistance change layer VRL and the metal layer M1. When On writing is performed from this state, these oxygen O reacts with the oxygen vacancies Vo in the conductive filament FI and the oxygen vacancies Vo disappear, and the conductive filament FI becomes thin or has a low density. It is easy to cause the On state to become unstable. Therefore, even in the case of such a structure, a remarkable effect can be obtained by applying the on-writing method as shown in the first to fifth embodiments.

  As described above, according to the ReRAM of the sixth embodiment, the resistance change has a structure in which the On state is easily destabilized, in other words, a structure in which the Off state is relatively more stable than the On state. By applying the On writing method as shown in Embodiment Modes 1 to 5 to the element VR, the effect of stabilizing the On state can be obtained more remarkably.

(Embodiment 7)
In the first to sixth embodiments described above, as shown in FIG. 2, a configuration in which the memory cell MC storing 1-bit information includes one resistance change element VR and one selection transistor TR will be described as an example. However, the method described in each embodiment can be applied to a ReRAM having a so-called cross-point configuration.

  FIG. 17 is a diagram showing an outline of a configuration example of a memory cell of a cross-point type ReRAM. As illustrated, the resistance change element VR is connected to the word line WL and the bit line BL without going through a switch. It is desirable that the non-linear resistance element NLR is connected in series with the resistance change element VR. Which of the metal layer M1 and the metal layer M2 in the resistance change element VR is connected to the bit line BL is not particularly limited, but the following description will be made assuming that the metal layer M1 is connected to the bit line BL.

  FIG. 18 is a diagram showing an outline of a configuration example of the memory cell array in the cross-point type ReRAM. The memory cell array MCA can be configured by arranging the memory cells MC shown in FIG. 17 in a matrix. The example of the memory cell array MCA shown in FIG. 18 has a 16-bit storage capacity composed of a matrix of 4 rows × 4 columns, but a larger storage capacity is realized by appropriately increasing the number of rows and columns in the array. be able to.

  Each memory cell MC is connected to each intersection of the word lines WL0 to WL3 and the bit lines BL0 to BL3. All word lines WL0 to WL3 and bit lines BL0 to BL3 are connected to a control circuit (not shown) at the periphery of the memory cell array MCA. For example, the word lines WL0 to WL3 are connected to a word line control circuit (not shown) on the left side of the figure in the memory cell array MCA. The bit lines BL0 to BL3 are connected to a bit line control circuit (not shown) in the upper part of the drawing.

  Each control circuit performs writing by appropriately applying a voltage to the bit line and the word line to bring a desired memory cell MC into a high resistance state or a low resistance state. Alternatively, reading is performed by detecting a current flowing through a bit line or a word line and determining whether a desired memory cell is in a high resistance state or a low resistance state.

  For example, in writing to turn on the memory cell MC surrounded by a dotted circle, the word line WL1 is set to a high potential, the bit line BL1 is set to a zero potential, and other word lines WL0, WL2, WL3, and bits The lines BL0, BL2, and BL3 may be set to ½ of the high potential. On the other hand, in writing in which the memory cell MC surrounded by a dotted circle is turned off, the word line WL1 is set to zero potential, the bit line BL1 is set to high potential, and the other word lines WL0, WL2, WL3, and The bit lines BL0, BL2, and BL3 may be set to ½ of the high potential.

  In addition, in order to read whether the memory cell MC surrounded by the dotted circle is on or off, the bit line BL1 is set to zero potential, the other bit lines BL0, BL2, BL3, and all the word lines WL0 to WL3. Is set to a high potential (but sufficiently lower than that at the time of writing), and the current flowing through the word line WL1 may be detected.

  As a result of the above operation, a high potential is applied to both ends only to the memory cells MC connected to the word line WL1 and the bit line BL1, and a ½ or zero potential of the high potential is applied to the other memory cells MC. As a result, only the memory cell MC surrounded by the dotted circle is written or read. The same applies when writing to or reading from other memory cells MC.

  Note that the non-linear resistance element NLR in the memory cell MC shown in FIG. 17 has a characteristic of high resistance when the potential difference between both ends is small and low resistance when the potential difference is large. Therefore, in FIG. 18, the memory cell MC surrounded by the dotted circle and the other memory cell MC sharing the bit line BL1 or the word line WL1, that is, a memory to which a voltage having a half of the high potential may be applied. The cell MC has a function of reducing the voltage applied to the resistance change element VR and preventing erroneous writing and erroneous reading.

  Even with the cross-point type ReRAM as described above, the method described in the first to sixth embodiments can be applied, and in a predetermined memory cell MC, the first time is short for the on-writing. After applying the On pulse, apply a pulse having the same polarity as that of the off write pulse (ie, opposite polarity to the On pulse) and a small absolute value, and then applying a longer On pulse. By taking the above, it is possible to improve the on-state retention characteristics while securing the memory window.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say.

VR resistance change element M1, M2 metal layer VRL resistance change layer MC memory cell WL word line BL bit line PL plate line TR selection transistor MCA memory cell array FI conductive filament NLR nonlinear resistance element

Claims (7)

A memory cell including a resistance change element;
A first write for applying a first write pulse to the memory cell in order to change the state of the memory cell to a first resistance state in which the resistance value of the variable resistance element is less than a first reference value. It is possible to perform a process and a second write process in which a second write pulse having a polarity opposite to that of the first write pulse is applied to obtain a second resistance state that is equal to or greater than a second reference value. A control circuit,
The control circuit has the same polarity as the first write pulse and a pulse from the first write pulse before applying the first write pulse to the memory cell in the first write process. A first pulse having a short width and a second pulse having the same polarity as the second write pulse are applied in this order ;
After the first write process, a read pulse is applied to read out whether the variable resistance element is in the first resistance state or the second resistance state, and the memory cell is in the first resistance state. If not, the second pulse is applied to the memory cell, the first write pulse is applied again, and then the read pulse is applied again so that the resistance change element A semiconductor memory device that performs a verify process for reading out whether the resistance state or the second resistance state . The semiconductor memory device according to claim 1,
The control circuit applies the set of the first pulse and the second pulse a plurality of times before applying the first write pulse to the memory cell in the first write process. Semiconductor memory device. The semiconductor memory device according to claim 1,
The semiconductor memory device, wherein the voltage of the first pulse is equal to or higher than the voltage of the first write pulse. The semiconductor memory device according to claim 1 ,
The control circuit repeats the verify process until the variable resistance element is in the first resistance state or until a predetermined upper limit number is reached. The semiconductor memory device according to claim 1,
In the second write process, the control circuit applies the second write pulse to the memory cell and then performs the second write process until the first write process is performed next. A semiconductor memory device that applies a third pulse having the same polarity as an address pulse and having a voltage lower than that of the second address pulse. The semiconductor memory device according to claim 1,
Of the two electrodes sandwiching the variable resistance layer forming the variable resistance element, the first metal layer constituting the electrode to which a positive voltage is applied when the first write process is performed is the variable resistance A semiconductor memory device including a metal having a property of being more easily reacted with oxygen than a material forming a layer. The semiconductor memory device according to claim 1,
Of the two electrodes sandwiching the variable resistance layer forming the variable resistance element, the second metal layer constituting the electrode to which a positive voltage is applied when performing the second write processing is a predetermined noble metal. A semiconductor memory device formed including the semiconductor device.